微创虚拟手术中的血管建模与碰撞检测算法研究

发布时间：2018-04-10 09:25

本文选题：微创虚拟血管手术　切入点：血管三维重建　出处：《大连理工大学》2015年硕士论文

【摘要】：微创虚拟血管手术仿真系统结合了计算机图形学、现代医学和虚拟仿真技术,通过对微创血管手术的模拟,为很多临床医生和实习生再现出了一个真实的手术场景,不但提高了用户的手术技能,同时大大减少手术误差。在保证安全的基础上,微创虚拟手术系统应为手术过程提供足够真实性的同时,还应使医生能自然和本能地控制工具的运动,充分利用其固有的感觉和反应能力,发挥其熟练手术技能。这就要求系统具有高临场感。为了解决仿真逼真性低、行为模型不够完善以及实时性和准确性相矛盾的问题,本文以腹主动脉血管为研究对象,对微创虚拟血管手术仿真系统中的三维模型重建、几何模型的优化、物理形变模型仿真、实时碰撞检测等关键技术进行了研究,研究主要内容概括如下：(1)血管的三维几何构建是微创虚拟血管手术系统的基础,本文针对血管造影图像对比度不均衡和血管边界模糊的问题,采用Mimics 15.0 (Materialise公司,比利时)软件对血管CT数据进行了分割、编辑和三维重建,同时对模型进行了相应的优化和检验。通过Mimics软件,采用混合自动分割的方式将血管组织分割出来,并通过空腔填充解决因造影图像对比度不均匀造成的血管蒙版信息丢失的问题,同时,通过蒙版编辑对血管进行边界增强并去噪。(2)针对血管壁复杂的组织构成,本文采用基于四面体网格的粘弹性有限元模型来模拟血管的形变。由于有限元模型在形变过程中需要进行大量的计算,在保证腹主动脉血管形变模型的精度的同时,结合真实血管的形变数据,本文将血管分为线性弹性形变部分和非线性形变部分。同时,本文利用Mimics软件对有限元的四面体网格进行分析和优化,获得高质量的体网格。(3)血管的有限元模型比较复杂,致使其碰撞检测计算量和计算时间相对较大。为了减少相交测试时间,提高检测精度,本文提出了基于轴向包围盒(Aligned Axis Bounding Box, AABB)和离散方向包围盒(Discrete Orientation Polytopes, K-DOPs)的混合双层包围盒碰撞检测算法。该方法分为粗糙检测和精确检测两个阶段,上层(外层)采用AABB快速排出不可能发生碰撞检测的几何集,下层(内层)采用K-DOPs较精确地测试可能发生碰撞的几何集并检测出碰撞点。本文通过模拟微创虚拟血管手术过程中血管与手术刀相接触的场景,对血管的形变模型和碰撞检测进行了检验。实验表明,通过Mimics进行三维重建和模型优化,不但提高了重建模型的精度,还极大地提高了建模的速度。改进的血管形变模型以及用Mimics优化后的体网格,不但减少了有限元形变求解计算量,同时较精确地描述了血管的形变模型。采用AABB-K-DOPs混合包围盒有效地减少了计算量,提高了碰撞检测速率。这为建立高临场感的微创虚拟手术系统奠定了坚实的基础。
[Abstract]:The minimally invasive virtual vascular surgery simulation system combines computer graphics, modern medicine and virtual simulation technology. Through the simulation of minimally invasive vascular surgery, it reproduces a real operation scene for many clinicians and interns.Not only improve the user's surgical skills, but also greatly reduce the surgical error.While ensuring safety, the minimally invasive virtual surgery system should provide sufficient authenticity for the surgical process, while also allowing doctors to control the movement of the tool naturally and instinctively and to take full advantage of their inherent senses and responsiveness,Give play to their skilled surgical skills.This requires that the system has a high sense of presence.In order to solve the problems of low lifelike simulation, imperfect behavior model and contradiction between real-time and accuracy, the three-dimensional model reconstruction in the minimally invasive virtual vascular surgery simulation system is studied in this paper.The key technologies such as geometric model optimization, physical deformation model simulation, real-time collision detection and so on are studied. The main contents of the research are summarized as follows: Three-dimensional geometry construction of blood vessel is the foundation of minimally invasive virtual vascular surgery system.In this paper, aiming at the problems of uneven contrast and blurred boundary of angiography images, Mimics 15.0 Materialise Corporation (Belgium) is used to segment, edit and reconstruct the CT data of blood vessels.At the same time, the model is optimized and tested.By using Mimics software, the vascular tissue is segmented by mixed automatic segmentation, and the problem of missing information of vascular mask caused by uneven contrast of angiography image is solved by cavity filling. At the same time,Aiming at the complex tissue structure of vascular wall, a viscoelastic finite element model based on tetrahedron mesh is used to simulate the deformation of blood vessel.Since the finite element model needs a lot of calculation in the process of deformation, the accuracy of the model is guaranteed, and the deformation data of the real blood vessel are combined.In this paper, blood vessels are divided into linear elastic deformation and nonlinear deformation.At the same time, using Mimics software to analyze and optimize the finite element tetrahedron mesh, the finite element model of high quality volume mesh.In order to reduce the intersecting test time and improve the detection accuracy, a hybrid double-layer bounding box collision detection algorithm based on axial bounding Axis Bounding box (AABB) and discrete bounding box concrete Orientation polytopes (K-DOPs) is proposed.The method can be divided into two stages: rough detection and accurate detection. The upper layer (outer layer) uses AABB to quickly remove the geometric sets which cannot be detected by collision.The lower layer (inner layer) uses K-DOPs to accurately test the geometric set of possible collisions and to detect collision points.In this paper, the deformation model and collision detection of blood vessels were tested by simulating the contact scene between blood vessels and scalpel during minimally invasive virtual vascular surgery.Experimental results show that 3D reconstruction and model optimization through Mimics not only improve the accuracy of the reconstruction model, but also greatly improve the speed of modeling.The improved vascular deformation model and the volume mesh optimized by Mimics not only reduce the computational complexity of finite element deformation solution, but also accurately describe the vascular deformation model.The AABB-K-DOPs hybrid bounding box can effectively reduce the computation and improve the collision detection rate.This has laid a solid foundation for the establishment of a high-sense minimally invasive virtual surgery system.
【学位授予单位】：大连理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：R654;TP391.9

【参考文献】